FT-IR and 1H NMR characterization of the products of an ...
11
Combustion and Flame 146 (2006) 52–62 www.elsevier.com/locate/combustflame FT-IR and 1 H NMR characterization of the products of an ethylene inverse diffusion flame Alexander Santamaría a , Fanor Mondragón a,∗ , Alejandro Molina b , Nathan D. Marsh c , Eric G. Eddings c , Adel F. Sarofim c a Institute of Chemistry, University of Antioquia, A.A. 1226, Medellín, Colombia b Combustion Research Facility, Sandia National Laboratories, Livermore, CA 94551, USA c Department of Chemical Engineering, University of Utah, Salt Lake City, UT 84112, USA Received 26 August 2005; received in revised form 25 March 2006; accepted 4 April 2006 Available online 30 May 2006 Abstract Knowledge of the chemical structure of young soot and its precursors is very useful in the understanding of the paths leading to soot particle inception. This paper presents analyses of the chemical functional groups, based on FT-IR and 1 H NMR spectroscopy of the products obtained in an ethylene inverse diffusion flame. The trends in the data indicate that the soluble fraction of the soot becomes progressively more aromatic and less aliphatic as the height above the burner increases. Results from 1 H NMR spectra of the chloroform-soluble soot samples taken at different heights above the burner corroborate the infrared results based on proton chemical shifts (Ha, Hα,Hβ , and Hγ ). The results indicate that the aliphatic β and γ hydrogens suffered the most drastic reduction, while the aromatic character increased considerably with height, particularly in the first half of the flame. © 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Soot; FT-IR; 1 H NMR; Inverse diffusion flame 1. Introduction The intensive use of fossil fuels, mainly in com- bustion processes, leads to the generation of a car- bonaceous material known as soot. Due to its high polyaromatic hydrocarbon content, this material even at low concentrations has serious consequences for human health, especially when the particulate matter is in the nanometer-sized range [1]. The health risks are not only mutagenic, but also potentially carcino- * Corresponding author. Fax: +57 4 210 6565. E-mail address: [email protected] (F. Mondragón). genic [1–4]. As a result, government agencies have introduced strict control measures that regulate the release of soot into the environment. Measures like these have encouraged research and development of environmentally friendly technologies [5–8]. The formation of soot is a very complex phenom- enon that involves homogeneous and heterogeneous processes and continuous competition between reac- tions [9–12]. Although the process of soot evolution has been investigated experimentally and theoreti- cally in many studies [8–44], there is still some infor- mation lacking on the chemical structure of the differ- ent compounds formed in the flame that lead to soot formation, especially during the early stages of soot 0010-2180/$ – see front matter © 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.combustflame.2006.04.008
Transcript of FT-IR and 1H NMR characterization of the products of an ...
Combustion and Flame 146 (2006) 52–62
www.elsevier.com/locate/combustflame
FT-IR and 1H NMR characterization of the products of an ethylene inverse diffusion flame
Alexander Santamaría a, Fanor Mondragón a,∗, Alejandro Molina b, Nathan D. Marsh c, Eric G. Eddings c, Adel F. Sarofim c
a Institute of Chemistry, University of Antioquia, A.A. 1226, Medellín, Colombia b Combustion Research Facility, Sandia National Laboratories, Livermore, CA 94551, USA c Department of Chemical Engineering, University of Utah, Salt Lake City, UT 84112, USA
Received 26 August 2005; received in revised form 25 March 2006; accepted 4 April 2006
Available online 30 May 2006
Abstract
Knowledge of the chemical structure of young soot and its precursors is very useful in the understanding of the paths leading to soot particle inception. This paper presents analyses of the chemical functional groups, based on FT-IR and 1H NMR spectroscopy of the products obtained in an ethylene inverse diffusion flame. The trends in the data indicate that the soluble fraction of the soot becomes progressively more aromatic and less aliphatic as the height above the burner increases. Results from 1H NMR spectra of the chloroform-soluble soot samples taken at different heights above the burner corroborate the infrared results based on proton chemical shifts (Ha, Hα, Hβ, and Hγ ). The results indicate that the aliphatic β and γ hydrogens suffered the most drastic reduction, while the aromatic character increased considerably with height, particularly in the first half of the flame. © 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
Keywords: Soot; FT-IR; 1H NMR; Inverse diffusion flame
1. Introduction
The intensive use of fossil fuels, mainly in com- bustion processes, leads to the generation of a car- bonaceous material known as soot. Due to its high polyaromatic hydrocarbon content, this material even at low concentrations has serious consequences for human health, especially when the particulate matter is in the nanometer-sized range [1]. The health risks are not only mutagenic, but also potentially carcino-
* Corresponding author. Fax: +57 4 210 6565. E-mail address: [email protected]
(F. Mondragón).
0010-2180/$ – see front matter © 2006 The Combustion Institute. doi:10.1016/j.combustflame.2006.04.008
genic [1–4]. As a result, government agencies have introduced strict control measures that regulate the release of soot into the environment. Measures like these have encouraged research and development of environmentally friendly technologies [5–8].
The formation of soot is a very complex phenom- enon that involves homogeneous and heterogeneous processes and continuous competition between reac- tions [9–12]. Although the process of soot evolution has been investigated experimentally and theoreti- cally in many studies [8–44], there is still some infor- mation lacking on the chemical structure of the differ- ent compounds formed in the flame that lead to soot formation, especially during the early stages of soot
Published by Elsevier Inc. All rights reserved.
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 53
inception. The acquisition of material with young soot characteristics represents an experimental challenge due to the fast oxidation and pyrolysis transforma- tions that take place in the flame, especially in typical combustion systems [12,13,23,24]. To study the early period of soot formation, several types of diffusion flames have been used extensively, and one of these is the coflow normal diffusion flame (NDF) [8,12,18– 21,33–35]. In the NDF, the fuel is introduced through the central tube of the burner while the oxidizer is in- troduced in an annular ring. An alternate approach is the inverse diffusion flame (IDF), which is basically obtained by inverting the air and fuel positions rela- tive to a NDF configuration; that is, the oxidizer goes through the central tube and the fuel is in the annular space [22–32,35].
Up to now, most of the studies that involve the early period of soot formation have been carried out in normal diffusion flames due to their simplicity. Un- fortunately, one of the disadvantages that this type of configuration presents is that soot particles and con- densed intermediate hydrocarbons are transported to- ward the centerline of the flame due to thermophoretic forces. Because these particles are formed on the in- ner or fuel-rich side of the reaction zone, they must pass through the main reaction zone prior to exiting the flame, thus exposing them to significant oxida- tion and carbonization processes [12,23,33]. There- fore, obtaining young soot from this configuration re- quires intrusive sampling into the center of the flame, which results in disturbance of the system. Needless to say, the amount of sample obtained under these conditions is often minimal.
Studies carried out by Dobbins and colleagues in a NDF identified for the first time the presence of young soot in combustion systems, using ther- mophoretic particle deposition on a cooled target [19– 21,33]. However, the results of this study were limited to transmission electron microscopy analysis due to the small amount of sample that could be collected. In another study using an underventilated normal diffu- sion flame configuration (U-NDF), a detailed analysis of young soot was obtained. Unfortunately, this type of configuration is somewhat difficult to operate in a reliable manner due to stability problems, which re- sulted in the flame being confined within the walls of a tube [34].
Although numerous investigations have been car- ried out on NDF configurations, very few have been done on an IDF. One of the first studies related to this type of configuration was published in 1955 by Arthur and Napier, who pointed out that the soot produced in an IDF was stickier and more viscous than the soot produced in a NDF [32]. However, it was only in 2002 that Blevins et al. confirmed that the soot generated in an IDF is chemically and morphologically similar
to the young soot found in a U-NDF, with the char- acteristic that it is 50% soluble in dichloromethane, allowing direct chemical analysis [22,23].
Another advantage of the IDF configuration com- pared with NDF is that it provides a better separa- tion of pyrolysis and oxidation processes. In addition, soot and intermediate products from these processes are transported away from the main reaction zone by thermophoretic forces, and since these products are formed on the outer or fuel-rich side of the flame, they do not pass through the main reaction zone and thus avoid significant oxidation or carbonization. This configuration provides an opportunity to gather large samples of young soot from the surroundings with- out need to invade the flame with the sampling probe. This approach opens the possibility of studying the early stages of soot formation in a direct way [22,23, 25,27,28].
Although many experimental and theoretical in- vestigations support the hypothesis that soot inception occurs through coagulation reactions of polycyclic aromatic hydrocarbons (PAHs) [10,13–17], which are formed quickly in the region of the flame just ahead of the nucleation zone (commonly defined as the on- set of the yellow luminosity in a flame), very little is known about the molecular structure of these com- pounds. In particular, information on the variation of the content of oxygen and alkyl groups in PAH along the flame is missing and few studies in this direction have been carried out [36–39].
Recently, Öktem et al. [39] used photoionization aerosol mass spectrometry (PIAMS) to characterize semivolatile material present on soot particles de- posited on an aluminum probe. The sample was col- lected at the soot inception region of a premixed eth- ylene flame. They found that at lower heights above the burner (below the soot inception point), the soot chemical composition was dominated by polycyclic aromatic hydrocarbons, whereas aliphatic compounds made a noticeable contribution to soot growth after the soot inception point, where the particles become older. Similar results were obtained by Ciajolo et al. [37,38] using FT-IR and UV–vis analyses. They found that the condensable material of soot collected in an ethylene premixed flame has a significant contribution of aliphatic hydrogens just after the soot inception point due to CH2 and CH3 groups. Although these studies were carried out on premixed flames, it should be noted that the current work reflects results from a diffusion flame.
One of the causes for the lack of structural infor- mation is that the spectroscopic characterization of these materials becomes difficult due to their low sol- ubility in solvents, and the amount of sample, which that is often minimal. Therefore, much of the infor- mation available for premixed and diffusion flames
54 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
comes from solid state NMR and MS analyses [39, 42–44]. However, it is difficult to obtain detailed structural information from soot taken from these flames since, in general, significant carbonization or dehydrogenation has taken place.
The goal of this work is to provide detailed chem- ical information on the structural changes in the solu- ble fraction of soot obtained in an ethylene IDF, using spectroscopic techniques such as FT-IR and nuclear magnetic resonance of liquids.
2. Experimental
2.1. Burner
The IDF burner consists of three concentric brass tubes, which are distributed as follows: a central tube or air jet (1.2 cm i.d.), an intermediate tube used for supplying the ethylene as fuel (3.8 cm i.d.), and an outer tube (7.5 cm i.d.) for a N2 stream that is used as a shield to prevent the formation of flames with the room air. The whole system is mounted on a screw- drive mechanism. The flow conditions for the gaseous system were set to 27, 117, and 289 cm3/s for air, fuel, and N2, respectively, so that the gas velocities at the burner mouth were 25 cm/s in all the experi- ments. Under these conditions the visible flame height was 60 mm and the maximum flame temperature was about 2100 K (see Fig. 1).
2.2. Sampling probe
Samples were taken at different heights on the lat- eral edge (6 mm radial position) from an ethylene IDF using a 10-cm-length stainless steel probe with a 1-mm-i.d. capillary tip. Initial experiments were car- ried out with a quartz probe to assess potential cat- alytic effects of the stainless steel probe, and results of both FT-IR and 1H NMR analyses were equiva- lent, regardless of the sampling probe. Since the probe is sampling from the periphery of the flame and was thus at a significantly lower temperature than if it were sampling along the centerline. Thus stainless steel probe was used for experimental convenience.
The probe was connected to a vacuum system in line with a Teflon filter of 0.25-µm pore diameter that was used to collect the particulate matter and low- molecular-weight compounds that condense at 40 C (the operating temperature of the heating tape). A cold trap was used to capture some of the volatile material that passed through the filter. The sampling time for each experiment depended on the amount of condens- able sample, which was different at each sampling position, and varied from 20 min to 1 h. Generally, sampling at low flame positions required much more time than at higher positions because the particulate matter at this point was just forming.
The samples thus collected were extracted with chloroform using an ultrasonic bath at room temper- ature for 15 min. The soluble fraction was separated
Fig. 1. Experimental setup of the IDF burner.
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 55
using a Teflon filter of 0.25-µm pore diameter. The chloroform was evaporated under an inert atmosphere using a vacuum stove at 60 C.
2.3. FT-IR analysis
FT-IR spectra were taken of the combined sam- ple, as well as of the chloroform soluble portion. It was observed that both spectra were very similar, which indicates that the IR signals are due predom- inantly to the structures present in the soluble mater- ial present in the soot particles. Therefore, the FT-IR analysis was carried out on the samples prior to chlo- roform extraction as described in the following para- graph.
A small amount of the total sample collected on the Teflon filter at heights of 6, 10, 15, 25, 30, 35, 45, and 60 mm and exhaust was taken to prepare a 1% KBr pellet. In each case, the sample and the KBr were dried at 60 C under nitrogen before preparing the pellet. Each spectrum was the result of a 300- scan accumulation, a value that provided the best sig- nal/noise ratio. A Nicolet Magna 560 spectrometer was used with a MCT/A detector operated in a wave- length range of 600–4000 cm−1. Each sample was taken from the flame at least thrice to estimate repro- ducibility. The uncertainty of the IR measurements was less than 5%.
Table 1 Working conditions for 1H NMR measurements
Frequency 300 MHz Spectral width 3900 Hz Memory capacity 32 K Pulse width 9.60 µs Acquisition time 1.0 s
2.4. 1H NMR analysis
1H NMR spectra of the chloroform soluble frac- tion of the samples collected on the filter at 6, 15, 35, and 45 mm and exhaust were taken. For the 1H NMR experiments, the extracts were redissolved in CDCl3 containing trace amounts of tetramethylsilane (TMS), which was used as an internal chemical shift reference to indicate in parts per million (ppm) the difference of the proton’s resonance frequency. This is a number that gives information of the chemical environment of the protons in the sample. All spectra were taken in a Bruker AMX 300 spectrometer equipped with a 5-mm-inner-diameter tube and operated under condi- tions specified in Table 1.
Each spectrum was integrated manually at least four times and the results were averaged to reduce the uncertainty (less than 5%) generated by the manual adjustment. Then the spectrum of each sample was separated into characteristic regions for different pro- tons (Ha, Hα, Hβ , and Hγ ) according to their position in the molecule, using as a guide the chemical shifts assigned to characterize heavy crude oil 1H NMR spectra [45–48]. See the corresponding 1H NMR as- signments in Table 2.
3. Results and discussion
3.1. FT-IR characterization
The infrared analysis was done on nine samples taken at different heights from the flame: 6, 10, 15, 25, 30, 35, 45, and 60 mm and exhaust. All FT-IR spectra showed C–H stretching features due to aro- matic, aliphatic, and acetylenic groups. Fig. 2a shows a typical infrared absorption spectrum of the mater- ial collected in the filter at a height of 6 mm on the
Table 2 Assignments of the proton chemical shifts in NMR spectra
Hydrogens Description δ (ppm)
Ha Aromatic hydrogens sterically hindered with bay, fjord regions and other geometrically similar zones.
6.5–9.0
Aromatic hydrogens in very peri-condensed PAHs. Aromatic hydrogens corresponding to a single aromatic ring.
Hα Aliphatic hydrogens in methylene groups α to two aromatic rings (fluorene type). 2.0–5.0 Aliphatic hydrogens in methylene groups α to an aromatic ring and β position to another (acenaphthene type). Aliphatic hydrogens in methyl groups placed in bay or fjord regions of polyaromatic hydrocarbons.
Hβ Alicyclic hydrogens in β position to two aromatic rings. 1.0–2.0 Aliphatic hydrogens in methyl or methylene groups in β position to an aromatic ring.
Hγ Aliphatic hydrogens in methyl groups γ or further to an aromatic ring. 0.5–1.0
(a)
(b)
Fig. 2. Infrared spectrum of soot taken (a) at a 6-mm height in an ethylene IDF burner and (b) at the exhaust.
lateral edge of an ethylene IDF (see Table 3 for peak assignments).
This spectrum shows a sharp peak around 3300 cm−1 assigned to acetylenic C–H stretching groups, whose presence can partly sustain the hypothesis pro- posed by Frenklach et al. about the growth of pol- yaromatic units through the HACA mechanism [10, 13–17]. This mechanism basically consists of two repetitive steps: hydrogen abstraction from the react- ing hydrocarbon by atomic hydrogen, followed by acetylene addition to the radical site formed. Under
these conditions, the hydrogen and acetylene concen- trations in the gas phase must be significant to be considered key species in the soot formation process. According to some previously published IDF experi- mental results [10,29,30], high concentrations of both hydrogen atoms and acetylene at the tip of the flame have been found; therefore, continuous alkylation re- actions can take place.
Even though the sample in Fig. 2a was taken very low in the flame, a well-defined aromatic C–H stretching is observed at 3030 cm−1, accompanied
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 57
Table 3 Assignment of infrared absorption peaks to functional groups
Absorption peaks (cm−1) Assignments
3300 Acetylenic C–H stretching 3030 Aromatic C–H stretching 2975–2925–2850 Asymmetric and symmetric C–H stretching of CH3, CH2 aliphatic groups, respectively 1720 Carbonyl C=O stretching 1595 Aromatic and alkene C=C stretching
1450–1380
{ Asymmetric CH3 and scissor CH2 deformations Symmetric CH3 deformation and cyclic CH2
1300–1000 C–C and C–H plane deformation of aromatic groups and ether C–O–C stretching 890–840–750 C–H bending out of the plane of condensed aromatic systems
Table 4 Relative intensities of infrared absorption peak heights of soot generated in an ethylene IDF as a function of height above the burner
Height (mm) Wavenumber (cm−1)
3300 3030 2975 2925 2850 1720 1620 880 840 740
6.0 0.422 0.388 0.533 1.000 0.529 0.656 0.357 0.393 0.696 0.758 10.0 0.608 0.934 0.556 1.000 0.533 0.603 0.801 0.963 1.896 2.050 15.0 0.726 1.046 0.588 1.000 0.593 0.336 0.778 1.082 2.042 2.085 25.0 1.129 1.603 0.671 1.000 0.553 0.028 1.802 3.000 4.612 4.235 30.0 1.076 1.675 0.640 1.000 0.614 0.0011 1.664 2.630 4.390 4.309 35.0 1.439 1.854 0.671 1.000 0.711 8.5E−4 2.471 2.844 4.561 4.464 45.0 1.404 2.350 0.723 1.000 0.733 5.3E−4 2.550 3.318 5.070 5.089 60.0 1.500 2.500 1.000 1.000 0.812 2.9E−4 2.812 4.875 7.250 7.375 Exhaust (200) 1.730 2.733 1.000 1.000 0.867 1.6E−4 3.526 6.333 7.667 8.200
Note. All data are normalized to the height of the 2925 cm−1 signal.
by the signals in the region from 1000 to 700 cm−1. This region has three well-defined peaks correspond- ing to C=C–H out-of-plane deformation due to mono-, di-, and trisubstituted aromatic compounds and peri/cata-condensed aromatic rings. Generally, the peri-condensed aromatic compounds show in- frared absorption peaks at 750, 840, and 850 cm−1, whereas the cata-condensed aromatic compounds show absorption peaks at 750 and 850 cm−1.
Concerning the presence of aliphatic structures, three characteristic sharp peaks assigned to aliphatic C–H stretching groups are seen at 2975, 2925, and 2850 cm−1, respectively. The intermediate peak (2925 cm−1) showed the most drastic change with the height of the flame. Therefore, it was used in the calculation of the relative intensities of the infrared absorption peak heights presented in Table 4, as was suggested by McKinnon et al. [36]. The presence of the C–H stretching in aliphatic groups comes mainly from methyl, methylene, and methine groups bonded to aromatic rings on PAHs or methylene bridges (flu- orene type) maintaining the interconnection of PAHs within a network [12,36–38].
Fig. 2a also shows other characteristic signals in the region between 1720 and 1300 cm−1, where the most important ones correspond to C=O stretching
of carboxylic acids at 1720 cm−1 [36,40,41], C=C stretching of aromatic or alkene groups that appear at 1595 cm−1, and aliphatic C–H plane deformations of CH2/CH3 groups at 1380 and 1450 cm−1, respec- tively.
Fig. 2b shows the spectrum of the sample taken at the exhaust, which in general displays the same ab- sorption peak pattern observed in Fig. 2a, with the exception of the signal at 1720 cm−1, which dis- appears almost completely. This change is accom- panied by the broadening of the region from 1000 to 1300 cm−1, which is very significant. This re- gion is a complex section of the infrared spectrum where signals corresponding to aromatic C–C and C–H plane deformation structures can appear, but the most important structures correspond to ether C–O–C stretching groups [40,41]. With the information cur- rently available, however, a detailed analysis cannot be done.
Table 4 shows the variation in the relative in- tensity peaks for each spectrum normalized by the 2925 cm−1 aliphatic signal. In this way, we avoid making absolute peak height comparisons between different spectra which can be affected by factors such as the thickness and concentration of the KBr pellet. The most significant change in the data reported in
58 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
Fig. 3. Ratio of the C–H aromatic, C–H acetylenic, and C=C aromatic signals to aliphatic C–H stretch (2925 cm−1) infrared absorption peak heights as a function of height above burner.
Table 4 was observed in the variation of the relative intensity of the aromatic (3030 cm−1) to aliphatic (2925 cm−1) C–H stretches as a function of height above the burner, whose ratio is plotted in Fig. 3. A significant increase of the aromatic and acetylenic relative ratio was observed during the first half on the flame, where there was a large temperature increase.
All samples showed the presence of carbonyl- type oxygenated functional groups (∼1720 cm−1), which were more significant at the lower flame po- sitions. The highest intensity of this signal was ob- served around 15 mm. Table 4 shows the variation of the ratio of carbonyl (1720 cm−1) to aliphatic (2925 cm−1) groups, where a drastic change in the signal at 1720 cm−1 appears at 25 mm, since the carbonyl group practically disappears. The origin of oxygenated functional groups in soot is not very clear and at the same time is not well documented in the scientific literature related to FT-IR analysis. Some authors have observed the presence of oxygenated groups in products of a laminar premixed flame by FT-IR (absorption peaks at 1700 and 3500 cm−1) [36,40,41]. However, they state that it is very diffi- cult to know whether these functionalities appear in situ or after sampling. In the present investigation, a clear trend with flame height was found, which sug- gests that these group of compounds may be involved in the soot formation process. However, one cannot rule out the possibility that the oxygen has been added in situ during sampling because the sampling probe is located near the interface of gaseous flows (air/fuel) where separation of the oxidation and carbonization processes is much more difficult to achieve, especially
near the base flame. However, to remove this type of potential artifact is also difficult.
3.2. 1H NMR characterization
1H NMR analysis is a technique routinely used to characterize gasoline, heavy crude oils, and solvent- soluble coal fractions. The hydrogen groups associ- ated with the chemical shift seen in 1H NMR spectra can yield structural information that allows classifica- tion of complex mixtures containing hundreds of aro- matic, naphthenic, paraffinic, olefinic, and isoparaf- finic compounds [45–48].
Fig. 4a shows the 1H NMR spectrum of the solu- ble fraction of soot obtained 6-mm above the burner. In the same figure, the different hydrogen types present in the sample are classified (see Table 2). The main assigned groups were hydrogen on aro- matic rings (Ha, 6.5–9.0 ppm); hydrogen of CH, CH2, CH3 groups on carbon atoms in α position to aromatic rings (Hα, 2.0–5.0 ppm); hydrogen of CH, CH2, CH3 groups on carbon atoms in β position to aromatic rings (Hβ , 1.0–2.0 ppm); and methyl hydrogen on carbons γ , δ, or greater position to aromatic rings, as well as methyl hydrogen of alkanes, cycloalkanes, and naphthenic rings (Hγ , 0.5–1.0 ppm).
More detailed information on the chemical shifts observed by 1H NMR is presented in Table 2. It is worth highlighting that the integration limits for hy- drogens that belong to CH, CH2, and CH3 groups in a determined region of the spectrum are not easy to establish due to the high degree of signal overlap.
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 59
Fig. 4. 1H NMR spectra of chloroform-soluble fraction of soot collected on the filter (a) at a 6-mm height in an ethylene IDF burner and (b) at the exhaust.
The sample collected at 6 mm was 94% soluble in chloroform. Its appearance was of an oily nature, very sticky and with a low softening point. At this point in the flame, there are gaseous species and tarry materi- als that can be identified as precursors of soot parti- cles. These products have a significant aliphatic com- ponent with predominance of methyl and methylenic groups on β and γ positions to aromatic rings. It is important to observe that there is a significant fraction of hydrogen on the γ position between 0.7 and 0.8 ppm, which suggests the existence of large aliphatic chains or saturated rings (naphthenic-type) joined to different types of aromatic rings. The region between 3.5 and 5.0 ppm can correspond to olefinic hydrogens or CH2 groups that connect two aromatic units (fluorene-type).
Fig. 4b shows the 1H NMR spectrum of the chlo- roform soluble fraction (∼49%) of soot obtained in the exhaust. When compared with Fig. 4a, Fig. 4b shows a remarkable decrease of all the hydrogen sig- nals of naphthenic, olefinic, and aliphatic character (chemical shift region between 0.5–6.0 ppm). This behavior corroborates the results obtained by FT-IR.
For comparative purposes, the region between 6.0 and 9.5 ppm of the spectra obtained at 6 mm and the exhaust was enlarged, as shown in Fig. 5. The chloroform-soluble fraction of the soot obtained in the exhaust shows a significant increase in the aro- matic character as well as an increase in the con- densation degree of aromatic units. Despite the com- plexity of the spectra, the degree of overlapping and
the difficulty to assign signals to specific structures, the peaks in this region have been assigned as fol- lows [45]: the region between 6.0 and 7.2 ppm has been assigned to hydrogens on a single aromatic ring; the region between 7.2 and 8.3 ppm has been as- signed to hydrogens on diaromatic and most triaro- matic and tetra-aromatic rings; the region between 8.3 and 8.9 ppm has been assigned to hydrogens on some tri- and tetra-aromatic rings; and the last region, 8.9– 9.5 ppm, is assigned to hydrogens on tetra-aromatic rings or higher. Fig. 5 also shows that the soot soluble fraction obtained at 6 mm contains the same type of aromatic structures as found in the exhaust, although in a smaller concentration and with a different distrib- ution. For example, in the exhaust the signal between 7.7 and 8.2 ppm is quite intense, indicating a greater amount of aromatic compounds of this type.
Fig. 6 shows a summary of the trends of each hy- drogen type observed by 1H NMR as a function of height above the burner. It can be observed that the aliphatic hydrogen fractions labeled as Hα, Hβ , and Hγ decrease with an increase in height above the burner, with the most drastic reduction evident with the β hydrogens (from 47.5 to 6.7%). This observa- tion is accompanied by an increase in the aromatic hydrogen content (from 28 to 85.9%). These results indicate that the observed aliphatic structures early in the flame undergo dealkylation and/or cyclization re- actions, leading to more compact structures.
The results described above seem to be contra- dictory to the results reported by Öktem et al. [39]
60 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
Fig. 5. Aromatic region enlargement of the 1H NMR spectra of the chloroform-soluble fraction of soot.
Fig. 6. Percentage variation of the different hydrogen types in the chloroform-soluble fraction of soot, based on the height above the burner.
and Ciajolo et al. [37,38]. However, this is not the case, since this study used a different flame configu- ration, where the carbonization and oxidation degrees are different from a premixed flame configuration. For a given nominal equivalent ratio, the soot col- lected from the periphery of an inverse diffusion flame (fuel-rich zone) would be less oxidized than the soot collected from a premixed flame. Thus, it is expected that soot and intermediate hydrocarbons produced in an inverse diffusion flame may have a different evo- lution history as height above the burner increases as compared to a premixed flame.
4. Concluding remarks
In the experiments reported in this work, FT-IR and 1H NMR have given useful information on the molecular structure of soot precursors as a function of height above the burner surface in an ethylene in- verse diffusion flame. Given that an IDF favors the collection of “young soot,” the results of this research suggest that the soot formation process involves sev- eral general steps in the gaseous phase as well as in the condensed phase, accompanied by competing re- actions among the original molecular fragments, tarry
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 61
materials and particles. The FT-IR spectra indicate that early in the flame the alkyl substituents are very important components of the tarry material. Similar results were obtained from 1H NMR spectra of the chloroform-soluble fraction of soot, which is of an oily nature, very sticky, and with a low softening point and can be associated with the tarry components of the flame. Given the large content of the aliphatic structures observed by 1H NMR in this fraction, it is very likely that, due to thermal effects, some of these aliphatic components may undergo bond scis- sion, as observed by the drastic reduction of the β
and γ hydrogens of CH2 and CH3 groups attached to aromatic rings. Such fragmentation produces more elemental molecular fragments, which may alkylate the tarry materials or the solid particles, and some of this material may rearrange to increase the quantity of aromatic components. The findings of this work may be useful in the development of theoretical soot evo- lution models, since many of the current models do not consider the contribution of the aliphatic species or decomposition reactions, which, according to the experimental results presented here, may be signifi- cant.
Acknowledgments
The authors thank the University of Antioquia for the financial support of the Sostenibilidad 2005 pro- gram. Useful comments of Mark Mikofski on the manuscript are acknowledged. A. Santamaría thanks Colciencias and the University of Antioquia for the Ph.D. scholarship and the Combustion Laboratory of the University of Utah for the visiting research grant in 2003.
References
[1] A.F. Sarofim, J.S. Lighty, E.G. Eddings, Prep. Symp. Am. Chem. Soc. Div. Fuel Chem. 47 (2) (2002) 618.
[2] C.J. Carrell, T.J. Carrell, H.L. Carrell, J.P. Glusker, E. Abu-Shaqara, C. Cortez, R.G. Harvey, Carcinogene- sis 15 (11) (1994) 2931–2936.
[3] G.L. Borosky, J. Org. Chem. 64 (21) (1999) 7738– 7744.
[4] Z. Zhou, Q. Dai, T. Gu, J. Chem. Inf. Comput. Sci. 43 (2) (2003) 615–621.
[5] H. Burtscher, Aerosol Sci. 36 (2005) 896–932. [6] D.B. Kittelson, Aerosol Sci. 29 (5–6) (1998) 575–588. [7] C.A. Miller, W.P. Linak, Primary Particles Generated
by the Combustion of Heavy Fuel Oil and Coal, Re- port No. EPA-600/R-02-093, National Risk Manage- ment Research Laboratory, 2002.
[8] J.B. Moss, C.D. Stewart, Fire Safe. J. 30 (1998) 229– 250.
[9] B.S. Haynes, in: W. Bartok, A.F. Sarofim (Eds.), Fossil Fuel Combustion, Wiley, New York, 1991, p. 261.
[10] M. Frenklach, Phys. Chem. Chem. Phys. 4 (11) (2002) 2028–2037.
[11] I.M. Kennedy, Prog. Energy Combust. Sci. 23 (1997) 95–132.
[12] K. Siegmann, K. Sattler, H.C. Siegmann, J. Electron Spectrosc. Relat. Phenom. 126 (2002) 191–202.
[13] A. Violi, G.A. Voth, A.F. Sarofim, Combust. Sci. Tech- nol. 176 (2004) 1–14.
[14] A. D’Anna, A. Violi, A. D’Alessio, A.F. Sarofim, Com- bust. Flame 127 (1–2) (2001) 1995–2003.
[15] A. Violi, A.F. Sarofim, T.N. Truong, Combust. Flame 126 (1–2) (2001) 1506–1515.
[16] A. Violi, Combust. Flame 139 (4) (2004) 279–287. [17] N.M. Marinov, W.J. Pitz, C.K. Westbrook, M.J.
Castaldi, S.M. Senkan, Combust. Sci. Technol. 116– 117 (1996) 211–287.
[18] F. Liu, H. Guo, G.J. Smallwood, O.L. Gulder, Combust. Theory Modelling 7 (2003) 301–315.
[19] R.A. Dobbins, R.A. Fletcher, H.C. Chang, Combust. Flame 115 (1998) 285–298.
[20] R.A. Dobbins, R.A. Fletcher, W. Lu, Combust. Flame 100 (1995) 301–309.
[21] R.A. Dobbins, H. Subremaniasivam, in: H. Bockhorn (Ed.), Soot Formation in Combustion, Mechanisms and Models, Springer-Verlag, Berlin, 1994, p. 290.
[22] L.G. Blevins, R.A. Fletcher, B.A. Benner, E.B. Steel, G.W. Mulholland, Proc. Combust. Inst. 29 (2002) 2325–2333.
[23] L.G. Blevins, N.Y. Yang, G.W. Mulholland, R.W. Davis, E.B. Steel, Prep. Symp. Am. Chem. Soc. Div. Fuel Chem. 47 (2) (2002) 740.
[24] V.R. Katta, L.G. Blevins, W.M. Roquemore, Combust. Flame 142 (2005) 33–51.
[25] C.R. Kaplan, K. Kailasanath, Combust. Flame 124 (1– 2) (2001) 275–294.
[26] D.B. Makel, I.M. Kennedy, Combust. Sci. Technol. 97 (1994) 303–314.
[27] G.W. Sidebotham, I. Glassman, Combust. Sci. Tech- nol. 81 (1992) 207–219.
[28] G.W. Sidebotham, I. Glassman, Combust. Flame 90 (1992) 269–283.
[29] K.T. Wu, R.H. Essenhigh, Proc. Combust. Inst. 20 (1984) 1925–1932.
[30] E.J. Lee, K.C. Oh, H.D. Shin, Fuel 84 (5) (2005) 543– 550.
[31] K.C. Oh, U.D. Lee, H.D. Shin, E.J. Lee, Combust. Flame 140 (2005) 249–254.
[32] J.R. Arthur, D.H. Napier, Proc. Combust. Inst. 5 (1955) 303–315.
[33] R.A. Dobbins, C.M. Megaridis, Langmuir 3 (1987) 254–259.
[34] S. Leonard, G.W. Mulholland, R. Puri, R.J. Santoro, Combust. Flame 98 (1994) 20–34.
[35] L.R. Vander Wal, Proc. Combust. Inst. 26 (1996) 2269– 2275.
[36] J.T. McKinnon, E. Meyer, J.B. Horward, Combust. Flame 105 (1996) 161–166.
[37] A. Ciajolo, B. Apicella, R. Barbella, A. Tregrossi, Proc. Combust. Inst. 27 (1998) 1481–1487.
62 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
[38] A. Ciajolo, R. Barbella, A. Tregrossi, L. Bonfanti, Combust. Sci. Technol. 153 (2000) 19–32.
[39] B. Öktem, M.P. Tolocka, B. Zhao, H. Wang, M.V. John- ston, Combust. Flame 142 (2005) 364–373.
[40] U. Kirchner, R. Vogt, C. Natzeckb, J. Goschnickb, Aerosol Sci. 34 (2003) 1323–1346.
[41] U. Kirchner, V. Scheer, R. Vogt, J. Phys. Chem. A 104 (2000) 8908–8915.
[42] R.J. Pugmire, M.S. Solum, Y.J. Jiang, A. F Sarofim, J. Veranth, H.H. Schobert, P.J. Pappano, Prep. Symp. Am. Chem. Soc. Div. Fuel Chem. 47 (2) (2002) 733– 735.
View publication statsView publication stats
[43] M.S. Solum, A.F. Sarofim, R.J. Pugmire, T.H. Fletcher, H. Zhang, Energy Fuels 15 (2001) 961–971.
[44] S. Yan, Y.J. Jiang, N.D. Marsh, E.G. Eddings, A.F. Sarofim, R.J. Pugmire, Energy Fuels 19 (2005) 1804– 1811.
[45] M.D. Guillen, C. Diaz, C.G. Blanco, Fuel Process. Technol. 58 (1998) 1–15.
[46] S.W. Lee, B. Glavincevski, Fuel Process. Technol. 60 (1999) 81–86.
[47] Y. Yang, B. Liu, H. Xi, X. Sun, T. Zhang, Fuel 82 (2003) 721–727.
[48] G.S. Kapur, S. Berger, Fuel 81 (2002) 883–892.
FT-IR and 1H NMR characterization of the products of an ethylene inverse diffusion flame
Alexander Santamaría a, Fanor Mondragón a,∗, Alejandro Molina b, Nathan D. Marsh c, Eric G. Eddings c, Adel F. Sarofim c
a Institute of Chemistry, University of Antioquia, A.A. 1226, Medellín, Colombia b Combustion Research Facility, Sandia National Laboratories, Livermore, CA 94551, USA c Department of Chemical Engineering, University of Utah, Salt Lake City, UT 84112, USA
Received 26 August 2005; received in revised form 25 March 2006; accepted 4 April 2006
Available online 30 May 2006
Abstract
Knowledge of the chemical structure of young soot and its precursors is very useful in the understanding of the paths leading to soot particle inception. This paper presents analyses of the chemical functional groups, based on FT-IR and 1H NMR spectroscopy of the products obtained in an ethylene inverse diffusion flame. The trends in the data indicate that the soluble fraction of the soot becomes progressively more aromatic and less aliphatic as the height above the burner increases. Results from 1H NMR spectra of the chloroform-soluble soot samples taken at different heights above the burner corroborate the infrared results based on proton chemical shifts (Ha, Hα, Hβ, and Hγ ). The results indicate that the aliphatic β and γ hydrogens suffered the most drastic reduction, while the aromatic character increased considerably with height, particularly in the first half of the flame. © 2006 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
Keywords: Soot; FT-IR; 1H NMR; Inverse diffusion flame
1. Introduction
The intensive use of fossil fuels, mainly in com- bustion processes, leads to the generation of a car- bonaceous material known as soot. Due to its high polyaromatic hydrocarbon content, this material even at low concentrations has serious consequences for human health, especially when the particulate matter is in the nanometer-sized range [1]. The health risks are not only mutagenic, but also potentially carcino-
* Corresponding author. Fax: +57 4 210 6565. E-mail address: [email protected]
(F. Mondragón).
0010-2180/$ – see front matter © 2006 The Combustion Institute. doi:10.1016/j.combustflame.2006.04.008
genic [1–4]. As a result, government agencies have introduced strict control measures that regulate the release of soot into the environment. Measures like these have encouraged research and development of environmentally friendly technologies [5–8].
The formation of soot is a very complex phenom- enon that involves homogeneous and heterogeneous processes and continuous competition between reac- tions [9–12]. Although the process of soot evolution has been investigated experimentally and theoreti- cally in many studies [8–44], there is still some infor- mation lacking on the chemical structure of the differ- ent compounds formed in the flame that lead to soot formation, especially during the early stages of soot
Published by Elsevier Inc. All rights reserved.
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 53
inception. The acquisition of material with young soot characteristics represents an experimental challenge due to the fast oxidation and pyrolysis transforma- tions that take place in the flame, especially in typical combustion systems [12,13,23,24]. To study the early period of soot formation, several types of diffusion flames have been used extensively, and one of these is the coflow normal diffusion flame (NDF) [8,12,18– 21,33–35]. In the NDF, the fuel is introduced through the central tube of the burner while the oxidizer is in- troduced in an annular ring. An alternate approach is the inverse diffusion flame (IDF), which is basically obtained by inverting the air and fuel positions rela- tive to a NDF configuration; that is, the oxidizer goes through the central tube and the fuel is in the annular space [22–32,35].
Up to now, most of the studies that involve the early period of soot formation have been carried out in normal diffusion flames due to their simplicity. Un- fortunately, one of the disadvantages that this type of configuration presents is that soot particles and con- densed intermediate hydrocarbons are transported to- ward the centerline of the flame due to thermophoretic forces. Because these particles are formed on the in- ner or fuel-rich side of the reaction zone, they must pass through the main reaction zone prior to exiting the flame, thus exposing them to significant oxida- tion and carbonization processes [12,23,33]. There- fore, obtaining young soot from this configuration re- quires intrusive sampling into the center of the flame, which results in disturbance of the system. Needless to say, the amount of sample obtained under these conditions is often minimal.
Studies carried out by Dobbins and colleagues in a NDF identified for the first time the presence of young soot in combustion systems, using ther- mophoretic particle deposition on a cooled target [19– 21,33]. However, the results of this study were limited to transmission electron microscopy analysis due to the small amount of sample that could be collected. In another study using an underventilated normal diffu- sion flame configuration (U-NDF), a detailed analysis of young soot was obtained. Unfortunately, this type of configuration is somewhat difficult to operate in a reliable manner due to stability problems, which re- sulted in the flame being confined within the walls of a tube [34].
Although numerous investigations have been car- ried out on NDF configurations, very few have been done on an IDF. One of the first studies related to this type of configuration was published in 1955 by Arthur and Napier, who pointed out that the soot produced in an IDF was stickier and more viscous than the soot produced in a NDF [32]. However, it was only in 2002 that Blevins et al. confirmed that the soot generated in an IDF is chemically and morphologically similar
to the young soot found in a U-NDF, with the char- acteristic that it is 50% soluble in dichloromethane, allowing direct chemical analysis [22,23].
Another advantage of the IDF configuration com- pared with NDF is that it provides a better separa- tion of pyrolysis and oxidation processes. In addition, soot and intermediate products from these processes are transported away from the main reaction zone by thermophoretic forces, and since these products are formed on the outer or fuel-rich side of the flame, they do not pass through the main reaction zone and thus avoid significant oxidation or carbonization. This configuration provides an opportunity to gather large samples of young soot from the surroundings with- out need to invade the flame with the sampling probe. This approach opens the possibility of studying the early stages of soot formation in a direct way [22,23, 25,27,28].
Although many experimental and theoretical in- vestigations support the hypothesis that soot inception occurs through coagulation reactions of polycyclic aromatic hydrocarbons (PAHs) [10,13–17], which are formed quickly in the region of the flame just ahead of the nucleation zone (commonly defined as the on- set of the yellow luminosity in a flame), very little is known about the molecular structure of these com- pounds. In particular, information on the variation of the content of oxygen and alkyl groups in PAH along the flame is missing and few studies in this direction have been carried out [36–39].
Recently, Öktem et al. [39] used photoionization aerosol mass spectrometry (PIAMS) to characterize semivolatile material present on soot particles de- posited on an aluminum probe. The sample was col- lected at the soot inception region of a premixed eth- ylene flame. They found that at lower heights above the burner (below the soot inception point), the soot chemical composition was dominated by polycyclic aromatic hydrocarbons, whereas aliphatic compounds made a noticeable contribution to soot growth after the soot inception point, where the particles become older. Similar results were obtained by Ciajolo et al. [37,38] using FT-IR and UV–vis analyses. They found that the condensable material of soot collected in an ethylene premixed flame has a significant contribution of aliphatic hydrogens just after the soot inception point due to CH2 and CH3 groups. Although these studies were carried out on premixed flames, it should be noted that the current work reflects results from a diffusion flame.
One of the causes for the lack of structural infor- mation is that the spectroscopic characterization of these materials becomes difficult due to their low sol- ubility in solvents, and the amount of sample, which that is often minimal. Therefore, much of the infor- mation available for premixed and diffusion flames
54 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
comes from solid state NMR and MS analyses [39, 42–44]. However, it is difficult to obtain detailed structural information from soot taken from these flames since, in general, significant carbonization or dehydrogenation has taken place.
The goal of this work is to provide detailed chem- ical information on the structural changes in the solu- ble fraction of soot obtained in an ethylene IDF, using spectroscopic techniques such as FT-IR and nuclear magnetic resonance of liquids.
2. Experimental
2.1. Burner
The IDF burner consists of three concentric brass tubes, which are distributed as follows: a central tube or air jet (1.2 cm i.d.), an intermediate tube used for supplying the ethylene as fuel (3.8 cm i.d.), and an outer tube (7.5 cm i.d.) for a N2 stream that is used as a shield to prevent the formation of flames with the room air. The whole system is mounted on a screw- drive mechanism. The flow conditions for the gaseous system were set to 27, 117, and 289 cm3/s for air, fuel, and N2, respectively, so that the gas velocities at the burner mouth were 25 cm/s in all the experi- ments. Under these conditions the visible flame height was 60 mm and the maximum flame temperature was about 2100 K (see Fig. 1).
2.2. Sampling probe
Samples were taken at different heights on the lat- eral edge (6 mm radial position) from an ethylene IDF using a 10-cm-length stainless steel probe with a 1-mm-i.d. capillary tip. Initial experiments were car- ried out with a quartz probe to assess potential cat- alytic effects of the stainless steel probe, and results of both FT-IR and 1H NMR analyses were equiva- lent, regardless of the sampling probe. Since the probe is sampling from the periphery of the flame and was thus at a significantly lower temperature than if it were sampling along the centerline. Thus stainless steel probe was used for experimental convenience.
The probe was connected to a vacuum system in line with a Teflon filter of 0.25-µm pore diameter that was used to collect the particulate matter and low- molecular-weight compounds that condense at 40 C (the operating temperature of the heating tape). A cold trap was used to capture some of the volatile material that passed through the filter. The sampling time for each experiment depended on the amount of condens- able sample, which was different at each sampling position, and varied from 20 min to 1 h. Generally, sampling at low flame positions required much more time than at higher positions because the particulate matter at this point was just forming.
The samples thus collected were extracted with chloroform using an ultrasonic bath at room temper- ature for 15 min. The soluble fraction was separated
Fig. 1. Experimental setup of the IDF burner.
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 55
using a Teflon filter of 0.25-µm pore diameter. The chloroform was evaporated under an inert atmosphere using a vacuum stove at 60 C.
2.3. FT-IR analysis
FT-IR spectra were taken of the combined sam- ple, as well as of the chloroform soluble portion. It was observed that both spectra were very similar, which indicates that the IR signals are due predom- inantly to the structures present in the soluble mater- ial present in the soot particles. Therefore, the FT-IR analysis was carried out on the samples prior to chlo- roform extraction as described in the following para- graph.
A small amount of the total sample collected on the Teflon filter at heights of 6, 10, 15, 25, 30, 35, 45, and 60 mm and exhaust was taken to prepare a 1% KBr pellet. In each case, the sample and the KBr were dried at 60 C under nitrogen before preparing the pellet. Each spectrum was the result of a 300- scan accumulation, a value that provided the best sig- nal/noise ratio. A Nicolet Magna 560 spectrometer was used with a MCT/A detector operated in a wave- length range of 600–4000 cm−1. Each sample was taken from the flame at least thrice to estimate repro- ducibility. The uncertainty of the IR measurements was less than 5%.
Table 1 Working conditions for 1H NMR measurements
Frequency 300 MHz Spectral width 3900 Hz Memory capacity 32 K Pulse width 9.60 µs Acquisition time 1.0 s
2.4. 1H NMR analysis
1H NMR spectra of the chloroform soluble frac- tion of the samples collected on the filter at 6, 15, 35, and 45 mm and exhaust were taken. For the 1H NMR experiments, the extracts were redissolved in CDCl3 containing trace amounts of tetramethylsilane (TMS), which was used as an internal chemical shift reference to indicate in parts per million (ppm) the difference of the proton’s resonance frequency. This is a number that gives information of the chemical environment of the protons in the sample. All spectra were taken in a Bruker AMX 300 spectrometer equipped with a 5-mm-inner-diameter tube and operated under condi- tions specified in Table 1.
Each spectrum was integrated manually at least four times and the results were averaged to reduce the uncertainty (less than 5%) generated by the manual adjustment. Then the spectrum of each sample was separated into characteristic regions for different pro- tons (Ha, Hα, Hβ , and Hγ ) according to their position in the molecule, using as a guide the chemical shifts assigned to characterize heavy crude oil 1H NMR spectra [45–48]. See the corresponding 1H NMR as- signments in Table 2.
3. Results and discussion
3.1. FT-IR characterization
The infrared analysis was done on nine samples taken at different heights from the flame: 6, 10, 15, 25, 30, 35, 45, and 60 mm and exhaust. All FT-IR spectra showed C–H stretching features due to aro- matic, aliphatic, and acetylenic groups. Fig. 2a shows a typical infrared absorption spectrum of the mater- ial collected in the filter at a height of 6 mm on the
Table 2 Assignments of the proton chemical shifts in NMR spectra
Hydrogens Description δ (ppm)
Ha Aromatic hydrogens sterically hindered with bay, fjord regions and other geometrically similar zones.
6.5–9.0
Aromatic hydrogens in very peri-condensed PAHs. Aromatic hydrogens corresponding to a single aromatic ring.
Hα Aliphatic hydrogens in methylene groups α to two aromatic rings (fluorene type). 2.0–5.0 Aliphatic hydrogens in methylene groups α to an aromatic ring and β position to another (acenaphthene type). Aliphatic hydrogens in methyl groups placed in bay or fjord regions of polyaromatic hydrocarbons.
Hβ Alicyclic hydrogens in β position to two aromatic rings. 1.0–2.0 Aliphatic hydrogens in methyl or methylene groups in β position to an aromatic ring.
Hγ Aliphatic hydrogens in methyl groups γ or further to an aromatic ring. 0.5–1.0
(a)
(b)
Fig. 2. Infrared spectrum of soot taken (a) at a 6-mm height in an ethylene IDF burner and (b) at the exhaust.
lateral edge of an ethylene IDF (see Table 3 for peak assignments).
This spectrum shows a sharp peak around 3300 cm−1 assigned to acetylenic C–H stretching groups, whose presence can partly sustain the hypothesis pro- posed by Frenklach et al. about the growth of pol- yaromatic units through the HACA mechanism [10, 13–17]. This mechanism basically consists of two repetitive steps: hydrogen abstraction from the react- ing hydrocarbon by atomic hydrogen, followed by acetylene addition to the radical site formed. Under
these conditions, the hydrogen and acetylene concen- trations in the gas phase must be significant to be considered key species in the soot formation process. According to some previously published IDF experi- mental results [10,29,30], high concentrations of both hydrogen atoms and acetylene at the tip of the flame have been found; therefore, continuous alkylation re- actions can take place.
Even though the sample in Fig. 2a was taken very low in the flame, a well-defined aromatic C–H stretching is observed at 3030 cm−1, accompanied
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 57
Table 3 Assignment of infrared absorption peaks to functional groups
Absorption peaks (cm−1) Assignments
3300 Acetylenic C–H stretching 3030 Aromatic C–H stretching 2975–2925–2850 Asymmetric and symmetric C–H stretching of CH3, CH2 aliphatic groups, respectively 1720 Carbonyl C=O stretching 1595 Aromatic and alkene C=C stretching
1450–1380
{ Asymmetric CH3 and scissor CH2 deformations Symmetric CH3 deformation and cyclic CH2
1300–1000 C–C and C–H plane deformation of aromatic groups and ether C–O–C stretching 890–840–750 C–H bending out of the plane of condensed aromatic systems
Table 4 Relative intensities of infrared absorption peak heights of soot generated in an ethylene IDF as a function of height above the burner
Height (mm) Wavenumber (cm−1)
3300 3030 2975 2925 2850 1720 1620 880 840 740
6.0 0.422 0.388 0.533 1.000 0.529 0.656 0.357 0.393 0.696 0.758 10.0 0.608 0.934 0.556 1.000 0.533 0.603 0.801 0.963 1.896 2.050 15.0 0.726 1.046 0.588 1.000 0.593 0.336 0.778 1.082 2.042 2.085 25.0 1.129 1.603 0.671 1.000 0.553 0.028 1.802 3.000 4.612 4.235 30.0 1.076 1.675 0.640 1.000 0.614 0.0011 1.664 2.630 4.390 4.309 35.0 1.439 1.854 0.671 1.000 0.711 8.5E−4 2.471 2.844 4.561 4.464 45.0 1.404 2.350 0.723 1.000 0.733 5.3E−4 2.550 3.318 5.070 5.089 60.0 1.500 2.500 1.000 1.000 0.812 2.9E−4 2.812 4.875 7.250 7.375 Exhaust (200) 1.730 2.733 1.000 1.000 0.867 1.6E−4 3.526 6.333 7.667 8.200
Note. All data are normalized to the height of the 2925 cm−1 signal.
by the signals in the region from 1000 to 700 cm−1. This region has three well-defined peaks correspond- ing to C=C–H out-of-plane deformation due to mono-, di-, and trisubstituted aromatic compounds and peri/cata-condensed aromatic rings. Generally, the peri-condensed aromatic compounds show in- frared absorption peaks at 750, 840, and 850 cm−1, whereas the cata-condensed aromatic compounds show absorption peaks at 750 and 850 cm−1.
Concerning the presence of aliphatic structures, three characteristic sharp peaks assigned to aliphatic C–H stretching groups are seen at 2975, 2925, and 2850 cm−1, respectively. The intermediate peak (2925 cm−1) showed the most drastic change with the height of the flame. Therefore, it was used in the calculation of the relative intensities of the infrared absorption peak heights presented in Table 4, as was suggested by McKinnon et al. [36]. The presence of the C–H stretching in aliphatic groups comes mainly from methyl, methylene, and methine groups bonded to aromatic rings on PAHs or methylene bridges (flu- orene type) maintaining the interconnection of PAHs within a network [12,36–38].
Fig. 2a also shows other characteristic signals in the region between 1720 and 1300 cm−1, where the most important ones correspond to C=O stretching
of carboxylic acids at 1720 cm−1 [36,40,41], C=C stretching of aromatic or alkene groups that appear at 1595 cm−1, and aliphatic C–H plane deformations of CH2/CH3 groups at 1380 and 1450 cm−1, respec- tively.
Fig. 2b shows the spectrum of the sample taken at the exhaust, which in general displays the same ab- sorption peak pattern observed in Fig. 2a, with the exception of the signal at 1720 cm−1, which dis- appears almost completely. This change is accom- panied by the broadening of the region from 1000 to 1300 cm−1, which is very significant. This re- gion is a complex section of the infrared spectrum where signals corresponding to aromatic C–C and C–H plane deformation structures can appear, but the most important structures correspond to ether C–O–C stretching groups [40,41]. With the information cur- rently available, however, a detailed analysis cannot be done.
Table 4 shows the variation in the relative in- tensity peaks for each spectrum normalized by the 2925 cm−1 aliphatic signal. In this way, we avoid making absolute peak height comparisons between different spectra which can be affected by factors such as the thickness and concentration of the KBr pellet. The most significant change in the data reported in
58 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
Fig. 3. Ratio of the C–H aromatic, C–H acetylenic, and C=C aromatic signals to aliphatic C–H stretch (2925 cm−1) infrared absorption peak heights as a function of height above burner.
Table 4 was observed in the variation of the relative intensity of the aromatic (3030 cm−1) to aliphatic (2925 cm−1) C–H stretches as a function of height above the burner, whose ratio is plotted in Fig. 3. A significant increase of the aromatic and acetylenic relative ratio was observed during the first half on the flame, where there was a large temperature increase.
All samples showed the presence of carbonyl- type oxygenated functional groups (∼1720 cm−1), which were more significant at the lower flame po- sitions. The highest intensity of this signal was ob- served around 15 mm. Table 4 shows the variation of the ratio of carbonyl (1720 cm−1) to aliphatic (2925 cm−1) groups, where a drastic change in the signal at 1720 cm−1 appears at 25 mm, since the carbonyl group practically disappears. The origin of oxygenated functional groups in soot is not very clear and at the same time is not well documented in the scientific literature related to FT-IR analysis. Some authors have observed the presence of oxygenated groups in products of a laminar premixed flame by FT-IR (absorption peaks at 1700 and 3500 cm−1) [36,40,41]. However, they state that it is very diffi- cult to know whether these functionalities appear in situ or after sampling. In the present investigation, a clear trend with flame height was found, which sug- gests that these group of compounds may be involved in the soot formation process. However, one cannot rule out the possibility that the oxygen has been added in situ during sampling because the sampling probe is located near the interface of gaseous flows (air/fuel) where separation of the oxidation and carbonization processes is much more difficult to achieve, especially
near the base flame. However, to remove this type of potential artifact is also difficult.
3.2. 1H NMR characterization
1H NMR analysis is a technique routinely used to characterize gasoline, heavy crude oils, and solvent- soluble coal fractions. The hydrogen groups associ- ated with the chemical shift seen in 1H NMR spectra can yield structural information that allows classifica- tion of complex mixtures containing hundreds of aro- matic, naphthenic, paraffinic, olefinic, and isoparaf- finic compounds [45–48].
Fig. 4a shows the 1H NMR spectrum of the solu- ble fraction of soot obtained 6-mm above the burner. In the same figure, the different hydrogen types present in the sample are classified (see Table 2). The main assigned groups were hydrogen on aro- matic rings (Ha, 6.5–9.0 ppm); hydrogen of CH, CH2, CH3 groups on carbon atoms in α position to aromatic rings (Hα, 2.0–5.0 ppm); hydrogen of CH, CH2, CH3 groups on carbon atoms in β position to aromatic rings (Hβ , 1.0–2.0 ppm); and methyl hydrogen on carbons γ , δ, or greater position to aromatic rings, as well as methyl hydrogen of alkanes, cycloalkanes, and naphthenic rings (Hγ , 0.5–1.0 ppm).
More detailed information on the chemical shifts observed by 1H NMR is presented in Table 2. It is worth highlighting that the integration limits for hy- drogens that belong to CH, CH2, and CH3 groups in a determined region of the spectrum are not easy to establish due to the high degree of signal overlap.
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 59
Fig. 4. 1H NMR spectra of chloroform-soluble fraction of soot collected on the filter (a) at a 6-mm height in an ethylene IDF burner and (b) at the exhaust.
The sample collected at 6 mm was 94% soluble in chloroform. Its appearance was of an oily nature, very sticky and with a low softening point. At this point in the flame, there are gaseous species and tarry materi- als that can be identified as precursors of soot parti- cles. These products have a significant aliphatic com- ponent with predominance of methyl and methylenic groups on β and γ positions to aromatic rings. It is important to observe that there is a significant fraction of hydrogen on the γ position between 0.7 and 0.8 ppm, which suggests the existence of large aliphatic chains or saturated rings (naphthenic-type) joined to different types of aromatic rings. The region between 3.5 and 5.0 ppm can correspond to olefinic hydrogens or CH2 groups that connect two aromatic units (fluorene-type).
Fig. 4b shows the 1H NMR spectrum of the chlo- roform soluble fraction (∼49%) of soot obtained in the exhaust. When compared with Fig. 4a, Fig. 4b shows a remarkable decrease of all the hydrogen sig- nals of naphthenic, olefinic, and aliphatic character (chemical shift region between 0.5–6.0 ppm). This behavior corroborates the results obtained by FT-IR.
For comparative purposes, the region between 6.0 and 9.5 ppm of the spectra obtained at 6 mm and the exhaust was enlarged, as shown in Fig. 5. The chloroform-soluble fraction of the soot obtained in the exhaust shows a significant increase in the aro- matic character as well as an increase in the con- densation degree of aromatic units. Despite the com- plexity of the spectra, the degree of overlapping and
the difficulty to assign signals to specific structures, the peaks in this region have been assigned as fol- lows [45]: the region between 6.0 and 7.2 ppm has been assigned to hydrogens on a single aromatic ring; the region between 7.2 and 8.3 ppm has been as- signed to hydrogens on diaromatic and most triaro- matic and tetra-aromatic rings; the region between 8.3 and 8.9 ppm has been assigned to hydrogens on some tri- and tetra-aromatic rings; and the last region, 8.9– 9.5 ppm, is assigned to hydrogens on tetra-aromatic rings or higher. Fig. 5 also shows that the soot soluble fraction obtained at 6 mm contains the same type of aromatic structures as found in the exhaust, although in a smaller concentration and with a different distrib- ution. For example, in the exhaust the signal between 7.7 and 8.2 ppm is quite intense, indicating a greater amount of aromatic compounds of this type.
Fig. 6 shows a summary of the trends of each hy- drogen type observed by 1H NMR as a function of height above the burner. It can be observed that the aliphatic hydrogen fractions labeled as Hα, Hβ , and Hγ decrease with an increase in height above the burner, with the most drastic reduction evident with the β hydrogens (from 47.5 to 6.7%). This observa- tion is accompanied by an increase in the aromatic hydrogen content (from 28 to 85.9%). These results indicate that the observed aliphatic structures early in the flame undergo dealkylation and/or cyclization re- actions, leading to more compact structures.
The results described above seem to be contra- dictory to the results reported by Öktem et al. [39]
60 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
Fig. 5. Aromatic region enlargement of the 1H NMR spectra of the chloroform-soluble fraction of soot.
Fig. 6. Percentage variation of the different hydrogen types in the chloroform-soluble fraction of soot, based on the height above the burner.
and Ciajolo et al. [37,38]. However, this is not the case, since this study used a different flame configu- ration, where the carbonization and oxidation degrees are different from a premixed flame configuration. For a given nominal equivalent ratio, the soot col- lected from the periphery of an inverse diffusion flame (fuel-rich zone) would be less oxidized than the soot collected from a premixed flame. Thus, it is expected that soot and intermediate hydrocarbons produced in an inverse diffusion flame may have a different evo- lution history as height above the burner increases as compared to a premixed flame.
4. Concluding remarks
In the experiments reported in this work, FT-IR and 1H NMR have given useful information on the molecular structure of soot precursors as a function of height above the burner surface in an ethylene in- verse diffusion flame. Given that an IDF favors the collection of “young soot,” the results of this research suggest that the soot formation process involves sev- eral general steps in the gaseous phase as well as in the condensed phase, accompanied by competing re- actions among the original molecular fragments, tarry
A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62 61
materials and particles. The FT-IR spectra indicate that early in the flame the alkyl substituents are very important components of the tarry material. Similar results were obtained from 1H NMR spectra of the chloroform-soluble fraction of soot, which is of an oily nature, very sticky, and with a low softening point and can be associated with the tarry components of the flame. Given the large content of the aliphatic structures observed by 1H NMR in this fraction, it is very likely that, due to thermal effects, some of these aliphatic components may undergo bond scis- sion, as observed by the drastic reduction of the β
and γ hydrogens of CH2 and CH3 groups attached to aromatic rings. Such fragmentation produces more elemental molecular fragments, which may alkylate the tarry materials or the solid particles, and some of this material may rearrange to increase the quantity of aromatic components. The findings of this work may be useful in the development of theoretical soot evo- lution models, since many of the current models do not consider the contribution of the aliphatic species or decomposition reactions, which, according to the experimental results presented here, may be signifi- cant.
Acknowledgments
The authors thank the University of Antioquia for the financial support of the Sostenibilidad 2005 pro- gram. Useful comments of Mark Mikofski on the manuscript are acknowledged. A. Santamaría thanks Colciencias and the University of Antioquia for the Ph.D. scholarship and the Combustion Laboratory of the University of Utah for the visiting research grant in 2003.
References
[1] A.F. Sarofim, J.S. Lighty, E.G. Eddings, Prep. Symp. Am. Chem. Soc. Div. Fuel Chem. 47 (2) (2002) 618.
[2] C.J. Carrell, T.J. Carrell, H.L. Carrell, J.P. Glusker, E. Abu-Shaqara, C. Cortez, R.G. Harvey, Carcinogene- sis 15 (11) (1994) 2931–2936.
[3] G.L. Borosky, J. Org. Chem. 64 (21) (1999) 7738– 7744.
[4] Z. Zhou, Q. Dai, T. Gu, J. Chem. Inf. Comput. Sci. 43 (2) (2003) 615–621.
[5] H. Burtscher, Aerosol Sci. 36 (2005) 896–932. [6] D.B. Kittelson, Aerosol Sci. 29 (5–6) (1998) 575–588. [7] C.A. Miller, W.P. Linak, Primary Particles Generated
by the Combustion of Heavy Fuel Oil and Coal, Re- port No. EPA-600/R-02-093, National Risk Manage- ment Research Laboratory, 2002.
[8] J.B. Moss, C.D. Stewart, Fire Safe. J. 30 (1998) 229– 250.
[9] B.S. Haynes, in: W. Bartok, A.F. Sarofim (Eds.), Fossil Fuel Combustion, Wiley, New York, 1991, p. 261.
[10] M. Frenklach, Phys. Chem. Chem. Phys. 4 (11) (2002) 2028–2037.
[11] I.M. Kennedy, Prog. Energy Combust. Sci. 23 (1997) 95–132.
[12] K. Siegmann, K. Sattler, H.C. Siegmann, J. Electron Spectrosc. Relat. Phenom. 126 (2002) 191–202.
[13] A. Violi, G.A. Voth, A.F. Sarofim, Combust. Sci. Tech- nol. 176 (2004) 1–14.
[14] A. D’Anna, A. Violi, A. D’Alessio, A.F. Sarofim, Com- bust. Flame 127 (1–2) (2001) 1995–2003.
[15] A. Violi, A.F. Sarofim, T.N. Truong, Combust. Flame 126 (1–2) (2001) 1506–1515.
[16] A. Violi, Combust. Flame 139 (4) (2004) 279–287. [17] N.M. Marinov, W.J. Pitz, C.K. Westbrook, M.J.
Castaldi, S.M. Senkan, Combust. Sci. Technol. 116– 117 (1996) 211–287.
[18] F. Liu, H. Guo, G.J. Smallwood, O.L. Gulder, Combust. Theory Modelling 7 (2003) 301–315.
[19] R.A. Dobbins, R.A. Fletcher, H.C. Chang, Combust. Flame 115 (1998) 285–298.
[20] R.A. Dobbins, R.A. Fletcher, W. Lu, Combust. Flame 100 (1995) 301–309.
[21] R.A. Dobbins, H. Subremaniasivam, in: H. Bockhorn (Ed.), Soot Formation in Combustion, Mechanisms and Models, Springer-Verlag, Berlin, 1994, p. 290.
[22] L.G. Blevins, R.A. Fletcher, B.A. Benner, E.B. Steel, G.W. Mulholland, Proc. Combust. Inst. 29 (2002) 2325–2333.
[23] L.G. Blevins, N.Y. Yang, G.W. Mulholland, R.W. Davis, E.B. Steel, Prep. Symp. Am. Chem. Soc. Div. Fuel Chem. 47 (2) (2002) 740.
[24] V.R. Katta, L.G. Blevins, W.M. Roquemore, Combust. Flame 142 (2005) 33–51.
[25] C.R. Kaplan, K. Kailasanath, Combust. Flame 124 (1– 2) (2001) 275–294.
[26] D.B. Makel, I.M. Kennedy, Combust. Sci. Technol. 97 (1994) 303–314.
[27] G.W. Sidebotham, I. Glassman, Combust. Sci. Tech- nol. 81 (1992) 207–219.
[28] G.W. Sidebotham, I. Glassman, Combust. Flame 90 (1992) 269–283.
[29] K.T. Wu, R.H. Essenhigh, Proc. Combust. Inst. 20 (1984) 1925–1932.
[30] E.J. Lee, K.C. Oh, H.D. Shin, Fuel 84 (5) (2005) 543– 550.
[31] K.C. Oh, U.D. Lee, H.D. Shin, E.J. Lee, Combust. Flame 140 (2005) 249–254.
[32] J.R. Arthur, D.H. Napier, Proc. Combust. Inst. 5 (1955) 303–315.
[33] R.A. Dobbins, C.M. Megaridis, Langmuir 3 (1987) 254–259.
[34] S. Leonard, G.W. Mulholland, R. Puri, R.J. Santoro, Combust. Flame 98 (1994) 20–34.
[35] L.R. Vander Wal, Proc. Combust. Inst. 26 (1996) 2269– 2275.
[36] J.T. McKinnon, E. Meyer, J.B. Horward, Combust. Flame 105 (1996) 161–166.
[37] A. Ciajolo, B. Apicella, R. Barbella, A. Tregrossi, Proc. Combust. Inst. 27 (1998) 1481–1487.
62 A. Santamaría et al. / Combustion and Flame 146 (2006) 52–62
[38] A. Ciajolo, R. Barbella, A. Tregrossi, L. Bonfanti, Combust. Sci. Technol. 153 (2000) 19–32.
[39] B. Öktem, M.P. Tolocka, B. Zhao, H. Wang, M.V. John- ston, Combust. Flame 142 (2005) 364–373.
[40] U. Kirchner, R. Vogt, C. Natzeckb, J. Goschnickb, Aerosol Sci. 34 (2003) 1323–1346.
[41] U. Kirchner, V. Scheer, R. Vogt, J. Phys. Chem. A 104 (2000) 8908–8915.
[42] R.J. Pugmire, M.S. Solum, Y.J. Jiang, A. F Sarofim, J. Veranth, H.H. Schobert, P.J. Pappano, Prep. Symp. Am. Chem. Soc. Div. Fuel Chem. 47 (2) (2002) 733– 735.
View publication statsView publication stats
[43] M.S. Solum, A.F. Sarofim, R.J. Pugmire, T.H. Fletcher, H. Zhang, Energy Fuels 15 (2001) 961–971.
[44] S. Yan, Y.J. Jiang, N.D. Marsh, E.G. Eddings, A.F. Sarofim, R.J. Pugmire, Energy Fuels 19 (2005) 1804– 1811.
[45] M.D. Guillen, C. Diaz, C.G. Blanco, Fuel Process. Technol. 58 (1998) 1–15.
[46] S.W. Lee, B. Glavincevski, Fuel Process. Technol. 60 (1999) 81–86.
[47] Y. Yang, B. Liu, H. Xi, X. Sun, T. Zhang, Fuel 82 (2003) 721–727.
[48] G.S. Kapur, S. Berger, Fuel 81 (2002) 883–892.
